As artificial intelligence is making its presence felt in many business sectors worldwide, some of its applications in Mexico are focused on social healthcare inclusion.
"The state of Jalisco has a very unique AI directorate," Enrique Cortes Rello, CEO of HealthCubed JCSA, told BNamericas. "The mandate is to work on the application of AI to social problems, and the priority is healthcare." Jalisco's AI directorate is an initiative of the state's innovation, science and technology department.
|
|
Cortes is also working with Jalisco's AI directorate to focus on the problem of diabetic retinopathy, a leading cause of blindness. "This is a well understood problem and there are open datasets."
Barriers to entry
The nature of data is one barrier to the rapid adoption of AI solutions in healthcare, regardless of the country. The requisite data may not exist, it may not apply to a certain group, or it may be so valuable that it is stored away.
Cortes demonstrates the point with a dataset of overweight female diabetes patients in the United States, which he uses in a data science class he teaches at the Tec de Monterrey university. "With that data set you are very unlikely to predict diabetes in the general population of Guadalajara."
While it is the best data available for students to learn data science concepts, it shows one limitation in delivering AI solutions. Furthermore, a shortage of technologists capable of using machine learning and other advanced quantitative skills is another barrier to scaling up AI solutions. And when it comes to medicine, AI is just beginning to make its way into the healthcare market because medical experts are justifiably wary of the "latest and greatest" technology until it is proven safe.
From research to startups
The impact felt so far from AI applications in Mexico originates from government centers and universities, as well as from abroad.
While there are dozens of healthtech startups in Mexico City, they tend to serve as platforms for booking doctor's appointments, online health payments and exercise monitoring - so there is a lack of firms looking to make a major healthcare impact.
 |
|
AI-powered medical startups may however be just around the corner and they would not be starting from scratch. "There are great frameworks that are relatively easy to use," said Cortes, citing IBM's Watson and Microsoft's Azure machine learning studio.
Meanwhile, a major regulatory hurdle toward the use of AI in healthcare globally is being overcome. "The FDA is gearing up to approve AI algorithms," said Cortes, who added that last year saw the US food and drug regulator give its first approval to an AI algorithm.
It is a mixed blessing. FDA approval means that an algorithm is no longer open-source, unlike the AI algorithm being used in Jalisco to detect diabetic retinopathy, which is based on open-source data. Still, it is helping to legitimate AI in healthcare, while also helping to standardize the process whereby AI algorithms will be assessed and approved by regulators in other nations.
Cortes said researchers are now working on prediction projects at Tec de Monterrey, Universidad de Guadalajara and Mexico's advanced research center (Cinvestav). Likewise, in other medical research institutes across Mexico, doctors and technologists are researching AI applications for nuclear medicine and molecular imaging.
Increasingly, the product of these research initiatives stands a chance of being taken up by startups, and introduced to the Mexican public.
Based on his experience teaching at Tec de Monterrey, Cortes said the university considers it "an important task to find and incentivize startups that use AI as a core technology in healthcare, and also to find social startups that could scale up using AI."
For now, Cortes is working with AI in a bid to detect diabetic retinopathy in Mexico but he expects the AI applications to expand into other areas soon. "The next set of use will have to do with prediction of chronic diseases."
Original article